]> git.baikalelectronics.ru Git - kernel.git/commit
mm: optimize PageHighMem() check
authorVineet Gupta <Vineet.Gupta1@synopsys.com>
Fri, 6 Nov 2015 02:48:29 +0000 (18:48 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Fri, 6 Nov 2015 03:34:48 +0000 (19:34 -0800)
commit1b0166dc4e07e7316c9441787d0174a6b5efa0e1
treef58acce501684287c2dd60f100afb43365ed000d
parent5bb472cb9d419b59f180ba380c2e70e02f72dff5
mm: optimize PageHighMem() check

This came up when implementing HIHGMEM/PAE40 for ARC.  The kmap() /
kmap_atomic() generated code seemed needlessly bloated due to the way
PageHighMem() macro is implemented.  It derives the exact zone for page
and then does pointer subtraction with first zone to infer the zone_type.
The pointer arithmatic in turn generates the code bloat.

PageHighMem(page)
  is_highmem(page_zone(page))
     zone_off = (char *)zone - (char *)zone->zone_pgdat->node_zones

Instead use is_highmem_idx() to work on zone_type available in page flags

   ----- Before -----
80756348: mov_s      r13,r0
8075634a: ld_s       r2,[r13,0]
8075634c: lsr_s      r2,r2,30
8075634e: mpy        r2,r2,0x2a4
80756352: add_s      r2,r2,0x80aef880
80756358: ld_s       r3,[r2,28]
8075635a: sub_s      r2,r2,r3
8075635c: breq       r2,0x2a4,80756378 <kmap+0x48>
80756364: breq       r2,0x548,80756378 <kmap+0x48>

   ----- After  -----
80756330: mov_s      r13,r0
80756332: ld_s       r2,[r13,0]
80756334: lsr_s      r2,r2,30
80756336: sub_s      r2,r2,1
80756338: brlo       r2,2,80756348 <kmap+0x30>

For x86 defconfig build (32 bit only) it saves around 900 bytes.
For ARC defconfig with HIGHMEM, it saved around 2K bytes.

   ---->8-------
./scripts/bloat-o-meter x86/vmlinux-defconfig-pre x86/vmlinux-defconfig-post
add/remove: 0/0 grow/shrink: 0/36 up/down: 0/-934 (-934)
function                                     old     new   delta
saveable_page                                162     154      -8
saveable_highmem_page                        154     146      -8
skb_gro_reset_offset                         147     131     -16
...
...
__change_page_attr_set_clr                  1715    1678     -37
setup_data_read                              434     394     -40
mon_bin_event                               1967    1927     -40
swsusp_save                                 1148    1105     -43
_set_pages_array                             549     493     -56
   ---->8-------

e.g. For ARC kmap()

Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Jennifer Herbert <jennifer.herbert@citrix.com>
Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
include/linux/page-flags.h